hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From mi...@apache.org
Subject [10/52] [partial] hbase-site git commit: Published site at 1cfd22bf43c9b64afae35d9bf16f764d0da80cab.
Date Tue, 21 Mar 2017 16:36:06 GMT
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
----------------------------------------------------------------------
diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
index a8ad4e5..4d16641 100644
--- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
+++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/LoadIncrementalHFiles.html
@@ -108,437 +108,437 @@
 <span class="sourceLineNo">100</span>import com.google.common.collect.Multimap;<a name="line.100"></a>
 <span class="sourceLineNo">101</span>import com.google.common.collect.Multimaps;<a name="line.101"></a>
 <span class="sourceLineNo">102</span>import com.google.common.util.concurrent.ThreadFactoryBuilder;<a name="line.102"></a>
-<span class="sourceLineNo">103</span><a name="line.103"></a>
-<span class="sourceLineNo">104</span>/**<a name="line.104"></a>
-<span class="sourceLineNo">105</span> * Tool to load the output of HFileOutputFormat into an existing table.<a name="line.105"></a>
-<span class="sourceLineNo">106</span> */<a name="line.106"></a>
-<span class="sourceLineNo">107</span>@InterfaceAudience.Public<a name="line.107"></a>
-<span class="sourceLineNo">108</span>@InterfaceStability.Stable<a name="line.108"></a>
-<span class="sourceLineNo">109</span>public class LoadIncrementalHFiles extends Configured implements Tool {<a name="line.109"></a>
-<span class="sourceLineNo">110</span>  private static final Log LOG = LogFactory.getLog(LoadIncrementalHFiles.class);<a name="line.110"></a>
-<span class="sourceLineNo">111</span>  private boolean initalized = false;<a name="line.111"></a>
-<span class="sourceLineNo">112</span><a name="line.112"></a>
-<span class="sourceLineNo">113</span>  public static final String NAME = "completebulkload";<a name="line.113"></a>
-<span class="sourceLineNo">114</span>  static final String RETRY_ON_IO_EXCEPTION = "hbase.bulkload.retries.retryOnIOException";<a name="line.114"></a>
-<span class="sourceLineNo">115</span>  public static final String MAX_FILES_PER_REGION_PER_FAMILY<a name="line.115"></a>
-<span class="sourceLineNo">116</span>    = "hbase.mapreduce.bulkload.max.hfiles.perRegion.perFamily";<a name="line.116"></a>
-<span class="sourceLineNo">117</span>  private static final String ASSIGN_SEQ_IDS = "hbase.mapreduce.bulkload.assign.sequenceNumbers";<a name="line.117"></a>
-<span class="sourceLineNo">118</span>  public final static String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.118"></a>
-<span class="sourceLineNo">119</span>  public final static String SILENCE_CONF_KEY = "ignore.unmatched.families";<a name="line.119"></a>
-<span class="sourceLineNo">120</span>  public final static String ALWAYS_COPY_FILES = "always.copy.files";<a name="line.120"></a>
-<span class="sourceLineNo">121</span><a name="line.121"></a>
-<span class="sourceLineNo">122</span>  // We use a '.' prefix which is ignored when walking directory trees<a name="line.122"></a>
-<span class="sourceLineNo">123</span>  // above. It is invalid family name.<a name="line.123"></a>
-<span class="sourceLineNo">124</span>  final static String TMP_DIR = ".tmp";<a name="line.124"></a>
-<span class="sourceLineNo">125</span><a name="line.125"></a>
-<span class="sourceLineNo">126</span>  private int maxFilesPerRegionPerFamily;<a name="line.126"></a>
-<span class="sourceLineNo">127</span>  private boolean assignSeqIds;<a name="line.127"></a>
-<span class="sourceLineNo">128</span>  private Set&lt;String&gt; unmatchedFamilies = new HashSet&lt;String&gt;();<a name="line.128"></a>
-<span class="sourceLineNo">129</span><a name="line.129"></a>
-<span class="sourceLineNo">130</span>  // Source filesystem<a name="line.130"></a>
-<span class="sourceLineNo">131</span>  private FileSystem fs;<a name="line.131"></a>
-<span class="sourceLineNo">132</span>  // Source delegation token<a name="line.132"></a>
-<span class="sourceLineNo">133</span>  private FsDelegationToken fsDelegationToken;<a name="line.133"></a>
-<span class="sourceLineNo">134</span>  private String bulkToken;<a name="line.134"></a>
-<span class="sourceLineNo">135</span>  private UserProvider userProvider;<a name="line.135"></a>
-<span class="sourceLineNo">136</span>  private int nrThreads;<a name="line.136"></a>
-<span class="sourceLineNo">137</span>  private RpcControllerFactory rpcControllerFactory;<a name="line.137"></a>
-<span class="sourceLineNo">138</span>  private AtomicInteger numRetries;<a name="line.138"></a>
-<span class="sourceLineNo">139</span><a name="line.139"></a>
-<span class="sourceLineNo">140</span>  private Map&lt;LoadQueueItem, ByteBuffer&gt; retValue = null;<a name="line.140"></a>
-<span class="sourceLineNo">141</span><a name="line.141"></a>
-<span class="sourceLineNo">142</span>  public LoadIncrementalHFiles(Configuration conf) throws Exception {<a name="line.142"></a>
-<span class="sourceLineNo">143</span>    super(conf);<a name="line.143"></a>
-<span class="sourceLineNo">144</span>    this.rpcControllerFactory = new RpcControllerFactory(conf);<a name="line.144"></a>
-<span class="sourceLineNo">145</span>    initialize();<a name="line.145"></a>
-<span class="sourceLineNo">146</span>  }<a name="line.146"></a>
-<span class="sourceLineNo">147</span><a name="line.147"></a>
-<span class="sourceLineNo">148</span>  private void initialize() throws Exception {<a name="line.148"></a>
-<span class="sourceLineNo">149</span>    if (initalized) {<a name="line.149"></a>
-<span class="sourceLineNo">150</span>      return;<a name="line.150"></a>
-<span class="sourceLineNo">151</span>    }<a name="line.151"></a>
-<span class="sourceLineNo">152</span>    // make a copy, just to be sure we're not overriding someone else's config<a name="line.152"></a>
-<span class="sourceLineNo">153</span>    setConf(HBaseConfiguration.create(getConf()));<a name="line.153"></a>
-<span class="sourceLineNo">154</span>    Configuration conf = getConf();<a name="line.154"></a>
-<span class="sourceLineNo">155</span>    // disable blockcache for tool invocation, see HBASE-10500<a name="line.155"></a>
-<span class="sourceLineNo">156</span>    conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.156"></a>
-<span class="sourceLineNo">157</span>    this.userProvider = UserProvider.instantiate(conf);<a name="line.157"></a>
-<span class="sourceLineNo">158</span>    this.fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.158"></a>
-<span class="sourceLineNo">159</span>    assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.159"></a>
-<span class="sourceLineNo">160</span>    maxFilesPerRegionPerFamily = conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.160"></a>
-<span class="sourceLineNo">161</span>    nrThreads = conf.getInt("hbase.loadincremental.threads.max",<a name="line.161"></a>
-<span class="sourceLineNo">162</span>      Runtime.getRuntime().availableProcessors());<a name="line.162"></a>
-<span class="sourceLineNo">163</span>    initalized = true;<a name="line.163"></a>
-<span class="sourceLineNo">164</span>    numRetries = new AtomicInteger(1);<a name="line.164"></a>
-<span class="sourceLineNo">165</span>  }<a name="line.165"></a>
-<span class="sourceLineNo">166</span><a name="line.166"></a>
-<span class="sourceLineNo">167</span>  private void usage() {<a name="line.167"></a>
-<span class="sourceLineNo">168</span>    System.err.println("usage: " + NAME + " /path/to/hfileoutputformat-output tablename" + "\n -D"<a name="line.168"></a>
-<span class="sourceLineNo">169</span>        + CREATE_TABLE_CONF_KEY + "=no - can be used to avoid creation of table by this tool\n"<a name="line.169"></a>
-<span class="sourceLineNo">170</span>        + "  Note: if you set this to 'no', then the target table must already exist in HBase\n -D"<a name="line.170"></a>
-<span class="sourceLineNo">171</span>        + SILENCE_CONF_KEY + "=yes - can be used to ignore unmatched column families\n"<a name="line.171"></a>
-<span class="sourceLineNo">172</span>        + "\n");<a name="line.172"></a>
-<span class="sourceLineNo">173</span>  }<a name="line.173"></a>
-<span class="sourceLineNo">174</span><a name="line.174"></a>
-<span class="sourceLineNo">175</span>  private interface BulkHFileVisitor&lt;TFamily&gt; {<a name="line.175"></a>
-<span class="sourceLineNo">176</span>    TFamily bulkFamily(final byte[] familyName)<a name="line.176"></a>
-<span class="sourceLineNo">177</span>      throws IOException;<a name="line.177"></a>
-<span class="sourceLineNo">178</span>    void bulkHFile(final TFamily family, final FileStatus hfileStatus)<a name="line.178"></a>
-<span class="sourceLineNo">179</span>      throws IOException;<a name="line.179"></a>
-<span class="sourceLineNo">180</span>  }<a name="line.180"></a>
-<span class="sourceLineNo">181</span><a name="line.181"></a>
-<span class="sourceLineNo">182</span>  /**<a name="line.182"></a>
-<span class="sourceLineNo">183</span>   * Iterate over the bulkDir hfiles.<a name="line.183"></a>
-<span class="sourceLineNo">184</span>   * Skip reference, HFileLink, files starting with "_" and non-valid hfiles.<a name="line.184"></a>
-<span class="sourceLineNo">185</span>   */<a name="line.185"></a>
-<span class="sourceLineNo">186</span>  private static &lt;TFamily&gt; void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.186"></a>
-<span class="sourceLineNo">187</span>    final BulkHFileVisitor&lt;TFamily&gt; visitor) throws IOException {<a name="line.187"></a>
-<span class="sourceLineNo">188</span>    visitBulkHFiles(fs, bulkDir, visitor, true);<a name="line.188"></a>
-<span class="sourceLineNo">189</span>  }<a name="line.189"></a>
-<span class="sourceLineNo">190</span><a name="line.190"></a>
-<span class="sourceLineNo">191</span>  /**<a name="line.191"></a>
-<span class="sourceLineNo">192</span>   * Iterate over the bulkDir hfiles.<a name="line.192"></a>
-<span class="sourceLineNo">193</span>   * Skip reference, HFileLink, files starting with "_".<a name="line.193"></a>
-<span class="sourceLineNo">194</span>   * Check and skip non-valid hfiles by default, or skip this validation by setting<a name="line.194"></a>
-<span class="sourceLineNo">195</span>   * 'hbase.loadincremental.validate.hfile' to false.<a name="line.195"></a>
-<span class="sourceLineNo">196</span>   */<a name="line.196"></a>
-<span class="sourceLineNo">197</span>  private static &lt;TFamily&gt; void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.197"></a>
-<span class="sourceLineNo">198</span>    final BulkHFileVisitor&lt;TFamily&gt; visitor, final boolean validateHFile) throws IOException {<a name="line.198"></a>
-<span class="sourceLineNo">199</span>    if (!fs.exists(bulkDir)) {<a name="line.199"></a>
-<span class="sourceLineNo">200</span>      throw new FileNotFoundException("Bulkload dir " + bulkDir + " not found");<a name="line.200"></a>
-<span class="sourceLineNo">201</span>    }<a name="line.201"></a>
-<span class="sourceLineNo">202</span><a name="line.202"></a>
-<span class="sourceLineNo">203</span>    FileStatus[] familyDirStatuses = fs.listStatus(bulkDir);<a name="line.203"></a>
-<span class="sourceLineNo">204</span>    if (familyDirStatuses == null) {<a name="line.204"></a>
-<span class="sourceLineNo">205</span>      throw new FileNotFoundException("No families found in " + bulkDir);<a name="line.205"></a>
-<span class="sourceLineNo">206</span>    }<a name="line.206"></a>
-<span class="sourceLineNo">207</span><a name="line.207"></a>
-<span class="sourceLineNo">208</span>    for (FileStatus familyStat : familyDirStatuses) {<a name="line.208"></a>
-<span class="sourceLineNo">209</span>      if (!familyStat.isDirectory()) {<a name="line.209"></a>
-<span class="sourceLineNo">210</span>        LOG.warn("Skipping non-directory " + familyStat.getPath());<a name="line.210"></a>
-<span class="sourceLineNo">211</span>        continue;<a name="line.211"></a>
-<span class="sourceLineNo">212</span>      }<a name="line.212"></a>
-<span class="sourceLineNo">213</span>      Path familyDir = familyStat.getPath();<a name="line.213"></a>
-<span class="sourceLineNo">214</span>      byte[] familyName = familyDir.getName().getBytes();<a name="line.214"></a>
-<span class="sourceLineNo">215</span>      // Skip invalid family<a name="line.215"></a>
-<span class="sourceLineNo">216</span>      try {<a name="line.216"></a>
-<span class="sourceLineNo">217</span>        HColumnDescriptor.isLegalFamilyName(familyName);<a name="line.217"></a>
-<span class="sourceLineNo">218</span>      }<a name="line.218"></a>
-<span class="sourceLineNo">219</span>      catch (IllegalArgumentException e) {<a name="line.219"></a>
-<span class="sourceLineNo">220</span>        LOG.warn("Skipping invalid " + familyStat.getPath());<a name="line.220"></a>
-<span class="sourceLineNo">221</span>        continue;<a name="line.221"></a>
-<span class="sourceLineNo">222</span>      }<a name="line.222"></a>
-<span class="sourceLineNo">223</span>      TFamily family = visitor.bulkFamily(familyName);<a name="line.223"></a>
-<span class="sourceLineNo">224</span><a name="line.224"></a>
-<span class="sourceLineNo">225</span>      FileStatus[] hfileStatuses = fs.listStatus(familyDir);<a name="line.225"></a>
-<span class="sourceLineNo">226</span>      for (FileStatus hfileStatus : hfileStatuses) {<a name="line.226"></a>
-<span class="sourceLineNo">227</span>        if (!fs.isFile(hfileStatus.getPath())) {<a name="line.227"></a>
-<span class="sourceLineNo">228</span>          LOG.warn("Skipping non-file " + hfileStatus);<a name="line.228"></a>
-<span class="sourceLineNo">229</span>          continue;<a name="line.229"></a>
-<span class="sourceLineNo">230</span>        }<a name="line.230"></a>
-<span class="sourceLineNo">231</span><a name="line.231"></a>
-<span class="sourceLineNo">232</span>        Path hfile = hfileStatus.getPath();<a name="line.232"></a>
-<span class="sourceLineNo">233</span>        // Skip "_", reference, HFileLink<a name="line.233"></a>
-<span class="sourceLineNo">234</span>        String fileName = hfile.getName();<a name="line.234"></a>
-<span class="sourceLineNo">235</span>        if (fileName.startsWith("_")) {<a name="line.235"></a>
-<span class="sourceLineNo">236</span>          continue;<a name="line.236"></a>
-<span class="sourceLineNo">237</span>        }<a name="line.237"></a>
-<span class="sourceLineNo">238</span>        if (StoreFileInfo.isReference(fileName)) {<a name="line.238"></a>
-<span class="sourceLineNo">239</span>          LOG.warn("Skipping reference " + fileName);<a name="line.239"></a>
-<span class="sourceLineNo">240</span>          continue;<a name="line.240"></a>
-<span class="sourceLineNo">241</span>        }<a name="line.241"></a>
-<span class="sourceLineNo">242</span>        if (HFileLink.isHFileLink(fileName)) {<a name="line.242"></a>
-<span class="sourceLineNo">243</span>          LOG.warn("Skipping HFileLink " + fileName);<a name="line.243"></a>
-<span class="sourceLineNo">244</span>          continue;<a name="line.244"></a>
-<span class="sourceLineNo">245</span>        }<a name="line.245"></a>
-<span class="sourceLineNo">246</span><a name="line.246"></a>
-<span class="sourceLineNo">247</span>        // Validate HFile Format if needed<a name="line.247"></a>
-<span class="sourceLineNo">248</span>        if (validateHFile) {<a name="line.248"></a>
-<span class="sourceLineNo">249</span>          try {<a name="line.249"></a>
-<span class="sourceLineNo">250</span>            if (!HFile.isHFileFormat(fs, hfile)) {<a name="line.250"></a>
-<span class="sourceLineNo">251</span>              LOG.warn("the file " + hfile + " doesn't seems to be an hfile. skipping");<a name="line.251"></a>
-<span class="sourceLineNo">252</span>              continue;<a name="line.252"></a>
-<span class="sourceLineNo">253</span>            }<a name="line.253"></a>
-<span class="sourceLineNo">254</span>          } catch (FileNotFoundException e) {<a name="line.254"></a>
-<span class="sourceLineNo">255</span>            LOG.warn("the file " + hfile + " was removed");<a name="line.255"></a>
-<span class="sourceLineNo">256</span>            continue;<a name="line.256"></a>
-<span class="sourceLineNo">257</span>          }<a name="line.257"></a>
-<span class="sourceLineNo">258</span>  }<a name="line.258"></a>
-<span class="sourceLineNo">259</span><a name="line.259"></a>
-<span class="sourceLineNo">260</span>        visitor.bulkHFile(family, hfileStatus);<a name="line.260"></a>
-<span class="sourceLineNo">261</span>      }<a name="line.261"></a>
-<span class="sourceLineNo">262</span>    }<a name="line.262"></a>
-<span class="sourceLineNo">263</span>  }<a name="line.263"></a>
-<span class="sourceLineNo">264</span><a name="line.264"></a>
-<span class="sourceLineNo">265</span>  /**<a name="line.265"></a>
-<span class="sourceLineNo">266</span>   * Represents an HFile waiting to be loaded. An queue is used<a name="line.266"></a>
-<span class="sourceLineNo">267</span>   * in this class in order to support the case where a region has<a name="line.267"></a>
-<span class="sourceLineNo">268</span>   * split during the process of the load. When this happens,<a name="line.268"></a>
-<span class="sourceLineNo">269</span>   * the HFile is split into two physical parts across the new<a name="line.269"></a>
-<span class="sourceLineNo">270</span>   * region boundary, and each part is added back into the queue.<a name="line.270"></a>
-<span class="sourceLineNo">271</span>   * The import process finishes when the queue is empty.<a name="line.271"></a>
-<span class="sourceLineNo">272</span>   */<a name="line.272"></a>
-<span class="sourceLineNo">273</span>  public static class LoadQueueItem {<a name="line.273"></a>
-<span class="sourceLineNo">274</span>    final byte[] family;<a name="line.274"></a>
-<span class="sourceLineNo">275</span>    final Path hfilePath;<a name="line.275"></a>
-<span class="sourceLineNo">276</span><a name="line.276"></a>
-<span class="sourceLineNo">277</span>    public LoadQueueItem(byte[] family, Path hfilePath) {<a name="line.277"></a>
-<span class="sourceLineNo">278</span>      this.family = family;<a name="line.278"></a>
-<span class="sourceLineNo">279</span>      this.hfilePath = hfilePath;<a name="line.279"></a>
-<span class="sourceLineNo">280</span>    }<a name="line.280"></a>
-<span class="sourceLineNo">281</span><a name="line.281"></a>
-<span class="sourceLineNo">282</span>    @Override<a name="line.282"></a>
-<span class="sourceLineNo">283</span>    public String toString() {<a name="line.283"></a>
-<span class="sourceLineNo">284</span>      return "family:"+ Bytes.toString(family) + " path:" + hfilePath.toString();<a name="line.284"></a>
-<span class="sourceLineNo">285</span>    }<a name="line.285"></a>
-<span class="sourceLineNo">286</span>  }<a name="line.286"></a>
-<span class="sourceLineNo">287</span><a name="line.287"></a>
-<span class="sourceLineNo">288</span>  /*<a name="line.288"></a>
-<span class="sourceLineNo">289</span>   * Populate the Queue with given HFiles<a name="line.289"></a>
-<span class="sourceLineNo">290</span>   */<a name="line.290"></a>
-<span class="sourceLineNo">291</span>  private void populateLoadQueue(final Deque&lt;LoadQueueItem&gt; ret,<a name="line.291"></a>
-<span class="sourceLineNo">292</span>      Map&lt;byte[], List&lt;Path&gt;&gt; map) throws IOException {<a name="line.292"></a>
-<span class="sourceLineNo">293</span>    for (Map.Entry&lt;byte[], List&lt;Path&gt;&gt; entry : map.entrySet()) {<a name="line.293"></a>
-<span class="sourceLineNo">294</span>      for (Path p : entry.getValue()) {<a name="line.294"></a>
-<span class="sourceLineNo">295</span>        ret.add(new LoadQueueItem(entry.getKey(), p));<a name="line.295"></a>
-<span class="sourceLineNo">296</span>      }<a name="line.296"></a>
-<span class="sourceLineNo">297</span>    }<a name="line.297"></a>
-<span class="sourceLineNo">298</span>  }<a name="line.298"></a>
-<span class="sourceLineNo">299</span><a name="line.299"></a>
-<span class="sourceLineNo">300</span>  /**<a name="line.300"></a>
-<span class="sourceLineNo">301</span>   * Walk the given directory for all HFiles, and return a Queue<a name="line.301"></a>
-<span class="sourceLineNo">302</span>   * containing all such files.<a name="line.302"></a>
-<span class="sourceLineNo">303</span>   */<a name="line.303"></a>
-<span class="sourceLineNo">304</span>  private void discoverLoadQueue(final Deque&lt;LoadQueueItem&gt; ret, final Path hfofDir,<a name="line.304"></a>
-<span class="sourceLineNo">305</span>      final boolean validateHFile) throws IOException {<a name="line.305"></a>
-<span class="sourceLineNo">306</span>    fs = hfofDir.getFileSystem(getConf());<a name="line.306"></a>
-<span class="sourceLineNo">307</span>    visitBulkHFiles(fs, hfofDir, new BulkHFileVisitor&lt;byte[]&gt;() {<a name="line.307"></a>
-<span class="sourceLineNo">308</span>      @Override<a name="line.308"></a>
-<span class="sourceLineNo">309</span>      public byte[] bulkFamily(final byte[] familyName) {<a name="line.309"></a>
-<span class="sourceLineNo">310</span>        return familyName;<a name="line.310"></a>
-<span class="sourceLineNo">311</span>      }<a name="line.311"></a>
-<span class="sourceLineNo">312</span>      @Override<a name="line.312"></a>
-<span class="sourceLineNo">313</span>      public void bulkHFile(final byte[] family, final FileStatus hfile) throws IOException {<a name="line.313"></a>
-<span class="sourceLineNo">314</span>        long length = hfile.getLen();<a name="line.314"></a>
-<span class="sourceLineNo">315</span>        if (length &gt; getConf().getLong(HConstants.HREGION_MAX_FILESIZE,<a name="line.315"></a>
-<span class="sourceLineNo">316</span>            HConstants.DEFAULT_MAX_FILE_SIZE)) {<a name="line.316"></a>
-<span class="sourceLineNo">317</span>          LOG.warn("Trying to bulk load hfile " + hfile.getPath() + " with size: " +<a name="line.317"></a>
-<span class="sourceLineNo">318</span>              length + " bytes can be problematic as it may lead to oversplitting.");<a name="line.318"></a>
-<span class="sourceLineNo">319</span>        }<a name="line.319"></a>
-<span class="sourceLineNo">320</span>        ret.add(new LoadQueueItem(family, hfile.getPath()));<a name="line.320"></a>
-<span class="sourceLineNo">321</span>      }<a name="line.321"></a>
-<span class="sourceLineNo">322</span>    }, validateHFile);<a name="line.322"></a>
-<span class="sourceLineNo">323</span>  }<a name="line.323"></a>
-<span class="sourceLineNo">324</span><a name="line.324"></a>
-<span class="sourceLineNo">325</span>  /**<a name="line.325"></a>
-<span class="sourceLineNo">326</span>   * Perform a bulk load of the given directory into the given<a name="line.326"></a>
-<span class="sourceLineNo">327</span>   * pre-existing table.  This method is not threadsafe.<a name="line.327"></a>
-<span class="sourceLineNo">328</span>   *<a name="line.328"></a>
-<span class="sourceLineNo">329</span>   * @param hfofDir the directory that was provided as the output path<a name="line.329"></a>
-<span class="sourceLineNo">330</span>   *   of a job using HFileOutputFormat<a name="line.330"></a>
-<span class="sourceLineNo">331</span>   * @param admin the Admin<a name="line.331"></a>
-<span class="sourceLineNo">332</span>   * @param table the table to load into<a name="line.332"></a>
-<span class="sourceLineNo">333</span>   * @param regionLocator region locator<a name="line.333"></a>
-<span class="sourceLineNo">334</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.334"></a>
-<span class="sourceLineNo">335</span>   */<a name="line.335"></a>
-<span class="sourceLineNo">336</span>  public void doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.336"></a>
-<span class="sourceLineNo">337</span>      RegionLocator regionLocator) throws TableNotFoundException, IOException {<a name="line.337"></a>
-<span class="sourceLineNo">338</span>    doBulkLoad(hfofDir, admin, table, regionLocator, false, false);<a name="line.338"></a>
-<span class="sourceLineNo">339</span>  }<a name="line.339"></a>
-<span class="sourceLineNo">340</span><a name="line.340"></a>
-<span class="sourceLineNo">341</span>  void cleanup(Admin admin, Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.341"></a>
-<span class="sourceLineNo">342</span>      SecureBulkLoadClient secureClient) throws IOException {<a name="line.342"></a>
-<span class="sourceLineNo">343</span>    fsDelegationToken.releaseDelegationToken();<a name="line.343"></a>
-<span class="sourceLineNo">344</span>    if (bulkToken != null &amp;&amp; secureClient != null) {<a name="line.344"></a>
-<span class="sourceLineNo">345</span>      secureClient.cleanupBulkLoad(admin.getConnection(), bulkToken);<a name="line.345"></a>
-<span class="sourceLineNo">346</span>    }<a name="line.346"></a>
-<span class="sourceLineNo">347</span>    if (pool != null) {<a name="line.347"></a>
-<span class="sourceLineNo">348</span>      pool.shutdown();<a name="line.348"></a>
-<span class="sourceLineNo">349</span>    }<a name="line.349"></a>
-<span class="sourceLineNo">350</span>    if (!queue.isEmpty()) {<a name="line.350"></a>
-<span class="sourceLineNo">351</span>      StringBuilder err = new StringBuilder();<a name="line.351"></a>
-<span class="sourceLineNo">352</span>      err.append("-------------------------------------------------\n");<a name="line.352"></a>
-<span class="sourceLineNo">353</span>      err.append("Bulk load aborted with some files not yet loaded:\n");<a name="line.353"></a>
-<span class="sourceLineNo">354</span>      err.append("-------------------------------------------------\n");<a name="line.354"></a>
-<span class="sourceLineNo">355</span>      for (LoadQueueItem q : queue) {<a name="line.355"></a>
-<span class="sourceLineNo">356</span>        err.append("  ").append(q.hfilePath).append('\n');<a name="line.356"></a>
-<span class="sourceLineNo">357</span>      }<a name="line.357"></a>
-<span class="sourceLineNo">358</span>      LOG.error(err);<a name="line.358"></a>
-<span class="sourceLineNo">359</span>    }<a name="line.359"></a>
-<span class="sourceLineNo">360</span>  }<a name="line.360"></a>
-<span class="sourceLineNo">361</span>  /**<a name="line.361"></a>
-<span class="sourceLineNo">362</span>   * Perform a bulk load of the given directory into the given<a name="line.362"></a>
-<span class="sourceLineNo">363</span>   * pre-existing table.  This method is not threadsafe.<a name="line.363"></a>
-<span class="sourceLineNo">364</span>   *<a name="line.364"></a>
-<span class="sourceLineNo">365</span>   * @param map map of family to List of hfiles<a name="line.365"></a>
-<span class="sourceLineNo">366</span>   * @param admin the Admin<a name="line.366"></a>
-<span class="sourceLineNo">367</span>   * @param table the table to load into<a name="line.367"></a>
-<span class="sourceLineNo">368</span>   * @param regionLocator region locator<a name="line.368"></a>
-<span class="sourceLineNo">369</span>   * @param silence true to ignore unmatched column families<a name="line.369"></a>
-<span class="sourceLineNo">370</span>   * @param copyFile always copy hfiles if true<a name="line.370"></a>
-<span class="sourceLineNo">371</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.371"></a>
-<span class="sourceLineNo">372</span>   */<a name="line.372"></a>
-<span class="sourceLineNo">373</span>  public void doBulkLoad(Map&lt;byte[], List&lt;Path&gt;&gt; map, final Admin admin,<a name="line.373"></a>
-<span class="sourceLineNo">374</span>      Table table, RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.374"></a>
-<span class="sourceLineNo">375</span>          throws TableNotFoundException, IOException {<a name="line.375"></a>
-<span class="sourceLineNo">376</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.376"></a>
-<span class="sourceLineNo">377</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.377"></a>
-<span class="sourceLineNo">378</span>    }<a name="line.378"></a>
-<span class="sourceLineNo">379</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.379"></a>
-<span class="sourceLineNo">380</span>    // happen in this thread<a name="line.380"></a>
-<span class="sourceLineNo">381</span>    Deque&lt;LoadQueueItem&gt; queue = new LinkedList&lt;&gt;();<a name="line.381"></a>
-<span class="sourceLineNo">382</span>    ExecutorService pool = null;<a name="line.382"></a>
-<span class="sourceLineNo">383</span>    SecureBulkLoadClient secureClient = null;<a name="line.383"></a>
-<span class="sourceLineNo">384</span>    try {<a name="line.384"></a>
-<span class="sourceLineNo">385</span>      prepareHFileQueue(map, table, queue, silence);<a name="line.385"></a>
-<span class="sourceLineNo">386</span>      if (queue.isEmpty()) {<a name="line.386"></a>
-<span class="sourceLineNo">387</span>        LOG.warn("Bulk load operation did not get any files to load");<a name="line.387"></a>
-<span class="sourceLineNo">388</span>        return;<a name="line.388"></a>
-<span class="sourceLineNo">389</span>      }<a name="line.389"></a>
-<span class="sourceLineNo">390</span>      pool = createExecutorService();<a name="line.390"></a>
-<span class="sourceLineNo">391</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.391"></a>
-<span class="sourceLineNo">392</span>      for (Map.Entry&lt;byte[], List&lt;Path&gt;&gt; entry : map.entrySet()) {<a name="line.392"></a>
-<span class="sourceLineNo">393</span>        for (Path p : entry.getValue()) {<a name="line.393"></a>
-<span class="sourceLineNo">394</span>          fs = p.getFileSystem(table.getConfiguration());<a name="line.394"></a>
-<span class="sourceLineNo">395</span>          break;<a name="line.395"></a>
-<span class="sourceLineNo">396</span>        }<a name="line.396"></a>
-<span class="sourceLineNo">397</span>      }<a name="line.397"></a>
-<span class="sourceLineNo">398</span>      retValue = performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.398"></a>
-<span class="sourceLineNo">399</span>    } finally {<a name="line.399"></a>
-<span class="sourceLineNo">400</span>      cleanup(admin, queue, pool, secureClient);<a name="line.400"></a>
-<span class="sourceLineNo">401</span>    }<a name="line.401"></a>
-<span class="sourceLineNo">402</span>  }<a name="line.402"></a>
-<span class="sourceLineNo">403</span><a name="line.403"></a>
-<span class="sourceLineNo">404</span>  /**<a name="line.404"></a>
-<span class="sourceLineNo">405</span>   * Perform a bulk load of the given directory into the given<a name="line.405"></a>
-<span class="sourceLineNo">406</span>   * pre-existing table.  This method is not threadsafe.<a name="line.406"></a>
-<span class="sourceLineNo">407</span>   *<a name="line.407"></a>
-<span class="sourceLineNo">408</span>   * @param hfofDir the directory that was provided as the output path<a name="line.408"></a>
-<span class="sourceLineNo">409</span>   *   of a job using HFileOutputFormat<a name="line.409"></a>
-<span class="sourceLineNo">410</span>   * @param admin the Admin<a name="line.410"></a>
-<span class="sourceLineNo">411</span>   * @param table the table to load into<a name="line.411"></a>
-<span class="sourceLineNo">412</span>   * @param regionLocator region locator<a name="line.412"></a>
-<span class="sourceLineNo">413</span>   * @param silence true to ignore unmatched column families<a name="line.413"></a>
-<span class="sourceLineNo">414</span>   * @param copyFile always copy hfiles if true<a name="line.414"></a>
-<span class="sourceLineNo">415</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.415"></a>
-<span class="sourceLineNo">416</span>   */<a name="line.416"></a>
-<span class="sourceLineNo">417</span>  public void doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.417"></a>
-<span class="sourceLineNo">418</span>      RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.418"></a>
-<span class="sourceLineNo">419</span>          throws TableNotFoundException, IOException {<a name="line.419"></a>
-<span class="sourceLineNo">420</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.420"></a>
-<span class="sourceLineNo">421</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.421"></a>
-<span class="sourceLineNo">422</span>    }<a name="line.422"></a>
-<span class="sourceLineNo">423</span><a name="line.423"></a>
-<span class="sourceLineNo">424</span>    /*<a name="line.424"></a>
-<span class="sourceLineNo">425</span>     * Checking hfile format is a time-consuming operation, we should have an option to skip<a name="line.425"></a>
-<span class="sourceLineNo">426</span>     * this step when bulkloading millions of HFiles. See HBASE-13985.<a name="line.426"></a>
-<span class="sourceLineNo">427</span>     */<a name="line.427"></a>
-<span class="sourceLineNo">428</span>    boolean validateHFile = getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a name="line.428"></a>
-<span class="sourceLineNo">429</span>    if (!validateHFile) {<a name="line.429"></a>
-<span class="sourceLineNo">430</span>      LOG.warn("You are skipping HFiles validation, it might cause some data loss if files " +<a name="line.430"></a>
-<span class="sourceLineNo">431</span>          "are not correct. If you fail to read data from your table after using this " +<a name="line.431"></a>
-<span class="sourceLineNo">432</span>          "option, consider removing the files and bulkload again without this option. " +<a name="line.432"></a>
-<span class="sourceLineNo">433</span>          "See HBASE-13985");<a name="line.433"></a>
-<span class="sourceLineNo">434</span>    }<a name="line.434"></a>
-<span class="sourceLineNo">435</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.435"></a>
-<span class="sourceLineNo">436</span>    // happen in this thread<a name="line.436"></a>
-<span class="sourceLineNo">437</span>    Deque&lt;LoadQueueItem&gt; queue = new LinkedList&lt;&gt;();<a name="line.437"></a>
-<span class="sourceLineNo">438</span>    ExecutorService pool = null;<a name="line.438"></a>
-<span class="sourceLineNo">439</span>    SecureBulkLoadClient secureClient = null;<a name="line.439"></a>
-<span class="sourceLineNo">440</span>    try {<a name="line.440"></a>
-<span class="sourceLineNo">441</span>      prepareHFileQueue(hfofDir, table, queue, validateHFile, silence);<a name="line.441"></a>
-<span class="sourceLineNo">442</span><a name="line.442"></a>
-<span class="sourceLineNo">443</span>      if (queue.isEmpty()) {<a name="line.443"></a>
-<span class="sourceLineNo">444</span>        LOG.warn("Bulk load operation did not find any files to load in " +<a name="line.444"></a>
-<span class="sourceLineNo">445</span>            "directory " + hfofDir != null ? hfofDir.toUri() : "" + ".  Does it contain files in " +<a name="line.445"></a>
-<span class="sourceLineNo">446</span>            "subdirectories that correspond to column family names?");<a name="line.446"></a>
-<span class="sourceLineNo">447</span>        return;<a name="line.447"></a>
-<span class="sourceLineNo">448</span>      }<a name="line.448"></a>
-<span class="sourceLineNo">449</span>      pool = createExecutorService();<a name="line.449"></a>
-<span class="sourceLineNo">450</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.450"></a>
-<span class="sourceLineNo">451</span>      retValue = performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.451"></a>
-<span class="sourceLineNo">452</span>    } finally {<a name="line.452"></a>
-<span class="sourceLineNo">453</span>      cleanup(admin, queue, pool, secureClient);<a name="line.453"></a>
-<span class="sourceLineNo">454</span>    }<a name="line.454"></a>
-<span class="sourceLineNo">455</span>  }<a name="line.455"></a>
-<span class="sourceLineNo">456</span><a name="line.456"></a>
-<span class="sourceLineNo">457</span>  Map&lt;LoadQueueItem, ByteBuffer&gt; performBulkLoad(final Admin admin, Table table,<a name="line.457"></a>
-<span class="sourceLineNo">458</span>      RegionLocator regionLocator, Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.458"></a>
-<span class="sourceLineNo">459</span>      SecureBulkLoadClient secureClient, boolean copyFile) throws IOException {<a name="line.459"></a>
-<span class="sourceLineNo">460</span>    int count = 0;<a name="line.460"></a>
-<span class="sourceLineNo">461</span><a name="line.461"></a>
-<span class="sourceLineNo">462</span>    if(isSecureBulkLoadEndpointAvailable()) {<a name="line.462"></a>
-<span class="sourceLineNo">463</span>      LOG.warn("SecureBulkLoadEndpoint is deprecated. It will be removed in future releases.");<a name="line.463"></a>
-<span class="sourceLineNo">464</span>      LOG.warn("Secure bulk load has been integrated into HBase core.");<a name="line.464"></a>
-<span class="sourceLineNo">465</span>    }<a name="line.465"></a>
-<span class="sourceLineNo">466</span><a name="line.466"></a>
-<span class="sourceLineNo">467</span>    //If using secure bulk load, get source delegation token, and<a name="line.467"></a>
-<span class="sourceLineNo">468</span>    //prepare staging directory and token<a name="line.468"></a>
-<span class="sourceLineNo">469</span>    // fs is the source filesystem<a name="line.469"></a>
-<span class="sourceLineNo">470</span>    fsDelegationToken.acquireDelegationToken(fs);<a name="line.470"></a>
-<span class="sourceLineNo">471</span>    bulkToken = secureClient.prepareBulkLoad(admin.getConnection());<a name="line.471"></a>
-<span class="sourceLineNo">472</span>    Pair&lt;Multimap&lt;ByteBuffer, LoadQueueItem&gt;, Set&lt;String&gt;&gt; pair = null;<a name="line.472"></a>
-<span class="sourceLineNo">473</span><a name="line.473"></a>
-<span class="sourceLineNo">474</span>    Map&lt;LoadQueueItem, ByteBuffer&gt; item2RegionMap = new HashMap&lt;&gt;();<a name="line.474"></a>
-<span class="sourceLineNo">475</span>    // Assumes that region splits can happen while this occurs.<a name="line.475"></a>
-<span class="sourceLineNo">476</span>    while (!queue.isEmpty()) {<a name="line.476"></a>
-<span class="sourceLineNo">477</span>      // need to reload split keys each iteration.<a name="line.477"></a>
-<span class="sourceLineNo">478</span>      final Pair&lt;byte[][], byte[][]&gt; startEndKeys = regionLocator.getStartEndKeys();<a name="line.478"></a>
-<span class="sourceLineNo">479</span>      if (count != 0) {<a name="line.479"></a>
-<span class="sourceLineNo">480</span>        LOG.info("Split occurred while grouping HFiles, retry attempt " +<a name="line.480"></a>
-<span class="sourceLineNo">481</span>            + count + " with " + queue.size() + " files remaining to group or split");<a name="line.481"></a>
-<span class="sourceLineNo">482</span>      }<a name="line.482"></a>
-<span class="sourceLineNo">483</span><a name="line.483"></a>
-<span class="sourceLineNo">484</span>      int maxRetries = getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.484"></a>
-<span class="sourceLineNo">485</span>      maxRetries = Math.max(maxRetries, startEndKeys.getFirst().length + 1);<a name="line.485"></a>
-<span class="sourceLineNo">486</span>      if (maxRetries != 0 &amp;&amp; count &gt;= maxRetries) {<a name="line.486"></a>
-<span class="sourceLineNo">487</span>        throw new IOException("Retry attempted " + count +<a name="line.487"></a>
-<span class="sourceLineNo">488</span>            " times without completing, bailing out");<a name="line.488"></a>
-<span class="sourceLineNo">489</span>      }<a name="line.489"></a>
-<span class="sourceLineNo">490</span>      count++;<a name="line.490"></a>
-<span class="sourceLineNo">491</span><a name="line.491"></a>
-<span class="sourceLineNo">492</span>      // Using ByteBuffer for byte[] equality semantics<a name="line.492"></a>
-<span class="sourceLineNo">493</span>      pair = groupOrSplitPhase(table, pool, queue, startEndKeys);<a name="line.493"></a>
-<span class="sourceLineNo">494</span>      Multimap&lt;ByteBuffer, LoadQueueItem&gt; regionGroups = pair.getFirst();<a name="line.494"></a>
-<span class="sourceLineNo">495</span><a name="line.495"></a>
-<span class="sourceLineNo">496</span>      if (!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.496"></a>
-<span class="sourceLineNo">497</span>        // Error is logged inside checkHFilesCountPerRegionPerFamily.<a name="line.497"></a>
-<span class="sourceLineNo">498</span>        throw new IOException("Trying to load more than " + maxFilesPerRegionPerFamily<a name="line.498"></a>
-<span class="sourceLineNo">499</span>            + " hfiles to one family of one region");<a name="line.499"></a>
-<span class="sourceLineNo">500</span>      }<a name="line.500"></a>
-<span class="sourceLineNo">501</span><a name="line.501"></a>
-<span class="sourceLineNo">502</span>      bulkLoadPhase(table, admin.getConnection(), pool, queue, regionGroups, copyFile,<a name="line.502"></a>
-<span class="sourceLineNo">503</span>          item2RegionMap);<a name="line.503"></a>
-<span class="sourceLineNo">504</span><a name="line.504"></a>
-<span class="sourceLineNo">505</span>      // NOTE: The next iteration's split / group could happen in parallel to<a name="line.505"></a>
-<span class="sourceLineNo">506</span>      // atomic bulkloads assuming that there are splits and no merges, and<a name="line.506"></a>
-<span class="sourceLineNo">507</span>      // that we can atomically pull out the groups we want to retry.<a name="line.507"></a>
-<span class="sourceLineNo">508</span>    }<a name="line.508"></a>
-<span class="sourceLineNo">509</span><a name="line.509"></a>
-<span class="sourceLineNo">510</span>    if (!queue.isEmpty()) {<a name="line.510"></a>
-<span class="sourceLineNo">511</span>      throw new RuntimeException("Bulk load aborted with some files not yet loaded."<a name="line.511"></a>
-<span class="sourceLineNo">512</span>        + "Please check log for more details.");<a name="line.512"></a>
-<span class="sourceLineNo">513</span>    }<a name="line.513"></a>
-<span class="sourceLineNo">514</span>    return item2RegionMap;<a name="line.514"></a>
-<span class="sourceLineNo">515</span>  }<a name="line.515"></a>
-<span class="sourceLineNo">516</span><a name="line.516"></a>
-<span class="sourceLineNo">517</span>  protected ClientServiceCallable&lt;byte[]&gt; buildClientServiceCallable(final Connection conn,<a name="line.517"></a>
-<span class="sourceLineNo">518</span>      TableName tableName, byte[] first, Collection&lt;LoadQueueItem&gt; lqis, boolean copyFile) {<a name="line.518"></a>
-<span class="sourceLineNo">519</span><a name="line.519"></a>
-<span class="sourceLineNo">520</span>    final List&lt;Pair&lt;byte[], String&gt;&gt; famPaths = new ArrayList&lt;&gt;(lqis.size());<a name="line.520"></a>
-<span class="sourceLineNo">521</span>    for (LoadQueueItem lqi : lqis) {<a name="line.521"></a>
-<span class="sourceLineNo">522</span>        famPaths.add(Pair.newPair(lqi.family, lqi.hfilePath.toString()));<a name="line.522"></a>
-<span class="sourceLineNo">523</span>    }<a name="line.523"></a>
-<span class="sourceLineNo">524</span><a name="line.524"></a>
-<span class="sourceLineNo">525</span>    return new ClientServiceCallable&lt;byte[]&gt;(conn,<a name="line.525"></a>
-<span class="sourceLineNo">526</span>        tableName, first, rpcControllerFactory.newController()) {<a name="line.526"></a>
-<span class="sourceLineNo">527</span>      @Override<a name="line.527"></a>
-<span class="sourceLineNo">528</span>      protected byte[] rpcCall() throws Exception {<a name="line.528"></a>
-<span class="sourceLineNo">529</span>        SecureBulkLoadClient secureClient = null;<a name="line.529"></a>
-<span class="sourceLineNo">530</span>        boolean success = false;<a name="line.530"></a>
-<span class="sourceLineNo">531</span>        try {<a name="line.531"></a>
-<span class="sourceLineNo">532</span>          LOG.debug("Going to connect to server " + getLocation() + " for row "<a name="line.532"></a>
-<span class="sourceLineNo">533</span>              + Bytes.toStringBinary(getRow()) + " with hfile group " + famPaths);<a name="line.533"></a>
+<span class="sourceLineNo">103</span>/**<a name="line.103"></a>
+<span class="sourceLineNo">104</span> * Tool to load the output of HFileOutputFormat into an existing table.<a name="line.104"></a>
+<span class="sourceLineNo">105</span> */<a name="line.105"></a>
+<span class="sourceLineNo">106</span>@InterfaceAudience.Public<a name="line.106"></a>
+<span class="sourceLineNo">107</span>@InterfaceStability.Stable<a name="line.107"></a>
+<span class="sourceLineNo">108</span>public class LoadIncrementalHFiles extends Configured implements Tool {<a name="line.108"></a>
+<span class="sourceLineNo">109</span>  private static final Log LOG = LogFactory.getLog(LoadIncrementalHFiles.class);<a name="line.109"></a>
+<span class="sourceLineNo">110</span>  private boolean initalized = false;<a name="line.110"></a>
+<span class="sourceLineNo">111</span><a name="line.111"></a>
+<span class="sourceLineNo">112</span>  public static final String NAME = "completebulkload";<a name="line.112"></a>
+<span class="sourceLineNo">113</span>  static final String RETRY_ON_IO_EXCEPTION = "hbase.bulkload.retries.retryOnIOException";<a name="line.113"></a>
+<span class="sourceLineNo">114</span>  public static final String MAX_FILES_PER_REGION_PER_FAMILY<a name="line.114"></a>
+<span class="sourceLineNo">115</span>    = "hbase.mapreduce.bulkload.max.hfiles.perRegion.perFamily";<a name="line.115"></a>
+<span class="sourceLineNo">116</span>  private static final String ASSIGN_SEQ_IDS = "hbase.mapreduce.bulkload.assign.sequenceNumbers";<a name="line.116"></a>
+<span class="sourceLineNo">117</span>  public final static String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.117"></a>
+<span class="sourceLineNo">118</span>  public final static String IGNORE_UNMATCHED_CF_CONF_KEY = "ignore.unmatched.families";<a name="line.118"></a>
+<span class="sourceLineNo">119</span>  public final static String ALWAYS_COPY_FILES = "always.copy.files";<a name="line.119"></a>
+<span class="sourceLineNo">120</span><a name="line.120"></a>
+<span class="sourceLineNo">121</span>  // We use a '.' prefix which is ignored when walking directory trees<a name="line.121"></a>
+<span class="sourceLineNo">122</span>  // above. It is invalid family name.<a name="line.122"></a>
+<span class="sourceLineNo">123</span>  final static String TMP_DIR = ".tmp";<a name="line.123"></a>
+<span class="sourceLineNo">124</span><a name="line.124"></a>
+<span class="sourceLineNo">125</span>  private int maxFilesPerRegionPerFamily;<a name="line.125"></a>
+<span class="sourceLineNo">126</span>  private boolean assignSeqIds;<a name="line.126"></a>
+<span class="sourceLineNo">127</span>  private Set&lt;String&gt; unmatchedFamilies = new HashSet&lt;&gt;();<a name="line.127"></a>
+<span class="sourceLineNo">128</span><a name="line.128"></a>
+<span class="sourceLineNo">129</span>  // Source filesystem<a name="line.129"></a>
+<span class="sourceLineNo">130</span>  private FileSystem fs;<a name="line.130"></a>
+<span class="sourceLineNo">131</span>  // Source delegation token<a name="line.131"></a>
+<span class="sourceLineNo">132</span>  private FsDelegationToken fsDelegationToken;<a name="line.132"></a>
+<span class="sourceLineNo">133</span>  private String bulkToken;<a name="line.133"></a>
+<span class="sourceLineNo">134</span>  private UserProvider userProvider;<a name="line.134"></a>
+<span class="sourceLineNo">135</span>  private int nrThreads;<a name="line.135"></a>
+<span class="sourceLineNo">136</span>  private RpcControllerFactory rpcControllerFactory;<a name="line.136"></a>
+<span class="sourceLineNo">137</span>  private AtomicInteger numRetries;<a name="line.137"></a>
+<span class="sourceLineNo">138</span><a name="line.138"></a>
+<span class="sourceLineNo">139</span>  private Map&lt;LoadQueueItem, ByteBuffer&gt; retValue = null;<a name="line.139"></a>
+<span class="sourceLineNo">140</span><a name="line.140"></a>
+<span class="sourceLineNo">141</span>  public LoadIncrementalHFiles(Configuration conf) throws Exception {<a name="line.141"></a>
+<span class="sourceLineNo">142</span>    super(conf);<a name="line.142"></a>
+<span class="sourceLineNo">143</span>    this.rpcControllerFactory = new RpcControllerFactory(conf);<a name="line.143"></a>
+<span class="sourceLineNo">144</span>    initialize();<a name="line.144"></a>
+<span class="sourceLineNo">145</span>  }<a name="line.145"></a>
+<span class="sourceLineNo">146</span><a name="line.146"></a>
+<span class="sourceLineNo">147</span>  private void initialize() throws Exception {<a name="line.147"></a>
+<span class="sourceLineNo">148</span>    if (initalized) {<a name="line.148"></a>
+<span class="sourceLineNo">149</span>      return;<a name="line.149"></a>
+<span class="sourceLineNo">150</span>    }<a name="line.150"></a>
+<span class="sourceLineNo">151</span>    // make a copy, just to be sure we're not overriding someone else's config<a name="line.151"></a>
+<span class="sourceLineNo">152</span>    setConf(HBaseConfiguration.create(getConf()));<a name="line.152"></a>
+<span class="sourceLineNo">153</span>    Configuration conf = getConf();<a name="line.153"></a>
+<span class="sourceLineNo">154</span>    // disable blockcache for tool invocation, see HBASE-10500<a name="line.154"></a>
+<span class="sourceLineNo">155</span>    conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0);<a name="line.155"></a>
+<span class="sourceLineNo">156</span>    this.userProvider = UserProvider.instantiate(conf);<a name="line.156"></a>
+<span class="sourceLineNo">157</span>    this.fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.157"></a>
+<span class="sourceLineNo">158</span>    assignSeqIds = conf.getBoolean(ASSIGN_SEQ_IDS, true);<a name="line.158"></a>
+<span class="sourceLineNo">159</span>    maxFilesPerRegionPerFamily = conf.getInt(MAX_FILES_PER_REGION_PER_FAMILY, 32);<a name="line.159"></a>
+<span class="sourceLineNo">160</span>    nrThreads = conf.getInt("hbase.loadincremental.threads.max",<a name="line.160"></a>
+<span class="sourceLineNo">161</span>      Runtime.getRuntime().availableProcessors());<a name="line.161"></a>
+<span class="sourceLineNo">162</span>    initalized = true;<a name="line.162"></a>
+<span class="sourceLineNo">163</span>    numRetries = new AtomicInteger(1);<a name="line.163"></a>
+<span class="sourceLineNo">164</span>  }<a name="line.164"></a>
+<span class="sourceLineNo">165</span><a name="line.165"></a>
+<span class="sourceLineNo">166</span>  private void usage() {<a name="line.166"></a>
+<span class="sourceLineNo">167</span>    System.err.println("usage: " + NAME + " /path/to/hfileoutputformat-output tablename" + "\n -D"<a name="line.167"></a>
+<span class="sourceLineNo">168</span>        + CREATE_TABLE_CONF_KEY + "=no - can be used to avoid creation of table by this tool\n"<a name="line.168"></a>
+<span class="sourceLineNo">169</span>        + "  Note: if you set this to 'no', then the target table must already exist in HBase\n -D"<a name="line.169"></a>
+<span class="sourceLineNo">170</span>        + IGNORE_UNMATCHED_CF_CONF_KEY + "=yes - can be used to ignore unmatched column families\n"<a name="line.170"></a>
+<span class="sourceLineNo">171</span>        + "\n");<a name="line.171"></a>
+<span class="sourceLineNo">172</span>  }<a name="line.172"></a>
+<span class="sourceLineNo">173</span><a name="line.173"></a>
+<span class="sourceLineNo">174</span>  private interface BulkHFileVisitor&lt;TFamily&gt; {<a name="line.174"></a>
+<span class="sourceLineNo">175</span>    TFamily bulkFamily(final byte[] familyName)<a name="line.175"></a>
+<span class="sourceLineNo">176</span>      throws IOException;<a name="line.176"></a>
+<span class="sourceLineNo">177</span>    void bulkHFile(final TFamily family, final FileStatus hfileStatus)<a name="line.177"></a>
+<span class="sourceLineNo">178</span>      throws IOException;<a name="line.178"></a>
+<span class="sourceLineNo">179</span>  }<a name="line.179"></a>
+<span class="sourceLineNo">180</span><a name="line.180"></a>
+<span class="sourceLineNo">181</span>  /**<a name="line.181"></a>
+<span class="sourceLineNo">182</span>   * Iterate over the bulkDir hfiles.<a name="line.182"></a>
+<span class="sourceLineNo">183</span>   * Skip reference, HFileLink, files starting with "_" and non-valid hfiles.<a name="line.183"></a>
+<span class="sourceLineNo">184</span>   */<a name="line.184"></a>
+<span class="sourceLineNo">185</span>  private static &lt;TFamily&gt; void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.185"></a>
+<span class="sourceLineNo">186</span>    final BulkHFileVisitor&lt;TFamily&gt; visitor) throws IOException {<a name="line.186"></a>
+<span class="sourceLineNo">187</span>    visitBulkHFiles(fs, bulkDir, visitor, true);<a name="line.187"></a>
+<span class="sourceLineNo">188</span>  }<a name="line.188"></a>
+<span class="sourceLineNo">189</span><a name="line.189"></a>
+<span class="sourceLineNo">190</span>  /**<a name="line.190"></a>
+<span class="sourceLineNo">191</span>   * Iterate over the bulkDir hfiles.<a name="line.191"></a>
+<span class="sourceLineNo">192</span>   * Skip reference, HFileLink, files starting with "_".<a name="line.192"></a>
+<span class="sourceLineNo">193</span>   * Check and skip non-valid hfiles by default, or skip this validation by setting<a name="line.193"></a>
+<span class="sourceLineNo">194</span>   * 'hbase.loadincremental.validate.hfile' to false.<a name="line.194"></a>
+<span class="sourceLineNo">195</span>   */<a name="line.195"></a>
+<span class="sourceLineNo">196</span>  private static &lt;TFamily&gt; void visitBulkHFiles(final FileSystem fs, final Path bulkDir,<a name="line.196"></a>
+<span class="sourceLineNo">197</span>    final BulkHFileVisitor&lt;TFamily&gt; visitor, final boolean validateHFile) throws IOException {<a name="line.197"></a>
+<span class="sourceLineNo">198</span>    if (!fs.exists(bulkDir)) {<a name="line.198"></a>
+<span class="sourceLineNo">199</span>      throw new FileNotFoundException("Bulkload dir " + bulkDir + " not found");<a name="line.199"></a>
+<span class="sourceLineNo">200</span>    }<a name="line.200"></a>
+<span class="sourceLineNo">201</span><a name="line.201"></a>
+<span class="sourceLineNo">202</span>    FileStatus[] familyDirStatuses = fs.listStatus(bulkDir);<a name="line.202"></a>
+<span class="sourceLineNo">203</span>    if (familyDirStatuses == null) {<a name="line.203"></a>
+<span class="sourceLineNo">204</span>      throw new FileNotFoundException("No families found in " + bulkDir);<a name="line.204"></a>
+<span class="sourceLineNo">205</span>    }<a name="line.205"></a>
+<span class="sourceLineNo">206</span><a name="line.206"></a>
+<span class="sourceLineNo">207</span>    for (FileStatus familyStat : familyDirStatuses) {<a name="line.207"></a>
+<span class="sourceLineNo">208</span>      if (!familyStat.isDirectory()) {<a name="line.208"></a>
+<span class="sourceLineNo">209</span>        LOG.warn("Skipping non-directory " + familyStat.getPath());<a name="line.209"></a>
+<span class="sourceLineNo">210</span>        continue;<a name="line.210"></a>
+<span class="sourceLineNo">211</span>      }<a name="line.211"></a>
+<span class="sourceLineNo">212</span>      Path familyDir = familyStat.getPath();<a name="line.212"></a>
+<span class="sourceLineNo">213</span>      byte[] familyName = familyDir.getName().getBytes();<a name="line.213"></a>
+<span class="sourceLineNo">214</span>      // Skip invalid family<a name="line.214"></a>
+<span class="sourceLineNo">215</span>      try {<a name="line.215"></a>
+<span class="sourceLineNo">216</span>        HColumnDescriptor.isLegalFamilyName(familyName);<a name="line.216"></a>
+<span class="sourceLineNo">217</span>      }<a name="line.217"></a>
+<span class="sourceLineNo">218</span>      catch (IllegalArgumentException e) {<a name="line.218"></a>
+<span class="sourceLineNo">219</span>        LOG.warn("Skipping invalid " + familyStat.getPath());<a name="line.219"></a>
+<span class="sourceLineNo">220</span>        continue;<a name="line.220"></a>
+<span class="sourceLineNo">221</span>      }<a name="line.221"></a>
+<span class="sourceLineNo">222</span>      TFamily family = visitor.bulkFamily(familyName);<a name="line.222"></a>
+<span class="sourceLineNo">223</span><a name="line.223"></a>
+<span class="sourceLineNo">224</span>      FileStatus[] hfileStatuses = fs.listStatus(familyDir);<a name="line.224"></a>
+<span class="sourceLineNo">225</span>      for (FileStatus hfileStatus : hfileStatuses) {<a name="line.225"></a>
+<span class="sourceLineNo">226</span>        if (!fs.isFile(hfileStatus.getPath())) {<a name="line.226"></a>
+<span class="sourceLineNo">227</span>          LOG.warn("Skipping non-file " + hfileStatus);<a name="line.227"></a>
+<span class="sourceLineNo">228</span>          continue;<a name="line.228"></a>
+<span class="sourceLineNo">229</span>        }<a name="line.229"></a>
+<span class="sourceLineNo">230</span><a name="line.230"></a>
+<span class="sourceLineNo">231</span>        Path hfile = hfileStatus.getPath();<a name="line.231"></a>
+<span class="sourceLineNo">232</span>        // Skip "_", reference, HFileLink<a name="line.232"></a>
+<span class="sourceLineNo">233</span>        String fileName = hfile.getName();<a name="line.233"></a>
+<span class="sourceLineNo">234</span>        if (fileName.startsWith("_")) {<a name="line.234"></a>
+<span class="sourceLineNo">235</span>          continue;<a name="line.235"></a>
+<span class="sourceLineNo">236</span>        }<a name="line.236"></a>
+<span class="sourceLineNo">237</span>        if (StoreFileInfo.isReference(fileName)) {<a name="line.237"></a>
+<span class="sourceLineNo">238</span>          LOG.warn("Skipping reference " + fileName);<a name="line.238"></a>
+<span class="sourceLineNo">239</span>          continue;<a name="line.239"></a>
+<span class="sourceLineNo">240</span>        }<a name="line.240"></a>
+<span class="sourceLineNo">241</span>        if (HFileLink.isHFileLink(fileName)) {<a name="line.241"></a>
+<span class="sourceLineNo">242</span>          LOG.warn("Skipping HFileLink " + fileName);<a name="line.242"></a>
+<span class="sourceLineNo">243</span>          continue;<a name="line.243"></a>
+<span class="sourceLineNo">244</span>        }<a name="line.244"></a>
+<span class="sourceLineNo">245</span><a name="line.245"></a>
+<span class="sourceLineNo">246</span>        // Validate HFile Format if needed<a name="line.246"></a>
+<span class="sourceLineNo">247</span>        if (validateHFile) {<a name="line.247"></a>
+<span class="sourceLineNo">248</span>          try {<a name="line.248"></a>
+<span class="sourceLineNo">249</span>            if (!HFile.isHFileFormat(fs, hfile)) {<a name="line.249"></a>
+<span class="sourceLineNo">250</span>              LOG.warn("the file " + hfile + " doesn't seems to be an hfile. skipping");<a name="line.250"></a>
+<span class="sourceLineNo">251</span>              continue;<a name="line.251"></a>
+<span class="sourceLineNo">252</span>            }<a name="line.252"></a>
+<span class="sourceLineNo">253</span>          } catch (FileNotFoundException e) {<a name="line.253"></a>
+<span class="sourceLineNo">254</span>            LOG.warn("the file " + hfile + " was removed");<a name="line.254"></a>
+<span class="sourceLineNo">255</span>            continue;<a name="line.255"></a>
+<span class="sourceLineNo">256</span>          }<a name="line.256"></a>
+<span class="sourceLineNo">257</span>  }<a name="line.257"></a>
+<span class="sourceLineNo">258</span><a name="line.258"></a>
+<span class="sourceLineNo">259</span>        visitor.bulkHFile(family, hfileStatus);<a name="line.259"></a>
+<span class="sourceLineNo">260</span>      }<a name="line.260"></a>
+<span class="sourceLineNo">261</span>    }<a name="line.261"></a>
+<span class="sourceLineNo">262</span>  }<a name="line.262"></a>
+<span class="sourceLineNo">263</span><a name="line.263"></a>
+<span class="sourceLineNo">264</span>  /**<a name="line.264"></a>
+<span class="sourceLineNo">265</span>   * Represents an HFile waiting to be loaded. An queue is used<a name="line.265"></a>
+<span class="sourceLineNo">266</span>   * in this class in order to support the case where a region has<a name="line.266"></a>
+<span class="sourceLineNo">267</span>   * split during the process of the load. When this happens,<a name="line.267"></a>
+<span class="sourceLineNo">268</span>   * the HFile is split into two physical parts across the new<a name="line.268"></a>
+<span class="sourceLineNo">269</span>   * region boundary, and each part is added back into the queue.<a name="line.269"></a>
+<span class="sourceLineNo">270</span>   * The import process finishes when the queue is empty.<a name="line.270"></a>
+<span class="sourceLineNo">271</span>   */<a name="line.271"></a>
+<span class="sourceLineNo">272</span>  public static class LoadQueueItem {<a name="line.272"></a>
+<span class="sourceLineNo">273</span>    final byte[] family;<a name="line.273"></a>
+<span class="sourceLineNo">274</span>    final Path hfilePath;<a name="line.274"></a>
+<span class="sourceLineNo">275</span><a name="line.275"></a>
+<span class="sourceLineNo">276</span>    public LoadQueueItem(byte[] family, Path hfilePath) {<a name="line.276"></a>
+<span class="sourceLineNo">277</span>      this.family = family;<a name="line.277"></a>
+<span class="sourceLineNo">278</span>      this.hfilePath = hfilePath;<a name="line.278"></a>
+<span class="sourceLineNo">279</span>    }<a name="line.279"></a>
+<span class="sourceLineNo">280</span><a name="line.280"></a>
+<span class="sourceLineNo">281</span>    @Override<a name="line.281"></a>
+<span class="sourceLineNo">282</span>    public String toString() {<a name="line.282"></a>
+<span class="sourceLineNo">283</span>      return "family:"+ Bytes.toString(family) + " path:" + hfilePath.toString();<a name="line.283"></a>
+<span class="sourceLineNo">284</span>    }<a name="line.284"></a>
+<span class="sourceLineNo">285</span>  }<a name="line.285"></a>
+<span class="sourceLineNo">286</span><a name="line.286"></a>
+<span class="sourceLineNo">287</span>  /*<a name="line.287"></a>
+<span class="sourceLineNo">288</span>   * Populate the Queue with given HFiles<a name="line.288"></a>
+<span class="sourceLineNo">289</span>   */<a name="line.289"></a>
+<span class="sourceLineNo">290</span>  private void populateLoadQueue(final Deque&lt;LoadQueueItem&gt; ret,<a name="line.290"></a>
+<span class="sourceLineNo">291</span>      Map&lt;byte[], List&lt;Path&gt;&gt; map) throws IOException {<a name="line.291"></a>
+<span class="sourceLineNo">292</span>    for (Map.Entry&lt;byte[], List&lt;Path&gt;&gt; entry : map.entrySet()) {<a name="line.292"></a>
+<span class="sourceLineNo">293</span>      for (Path p : entry.getValue()) {<a name="line.293"></a>
+<span class="sourceLineNo">294</span>        ret.add(new LoadQueueItem(entry.getKey(), p));<a name="line.294"></a>
+<span class="sourceLineNo">295</span>      }<a name="line.295"></a>
+<span class="sourceLineNo">296</span>    }<a name="line.296"></a>
+<span class="sourceLineNo">297</span>  }<a name="line.297"></a>
+<span class="sourceLineNo">298</span><a name="line.298"></a>
+<span class="sourceLineNo">299</span>  /**<a name="line.299"></a>
+<span class="sourceLineNo">300</span>   * Walk the given directory for all HFiles, and return a Queue<a name="line.300"></a>
+<span class="sourceLineNo">301</span>   * containing all such files.<a name="line.301"></a>
+<span class="sourceLineNo">302</span>   */<a name="line.302"></a>
+<span class="sourceLineNo">303</span>  private void discoverLoadQueue(final Deque&lt;LoadQueueItem&gt; ret, final Path hfofDir,<a name="line.303"></a>
+<span class="sourceLineNo">304</span>      final boolean validateHFile) throws IOException {<a name="line.304"></a>
+<span class="sourceLineNo">305</span>    fs = hfofDir.getFileSystem(getConf());<a name="line.305"></a>
+<span class="sourceLineNo">306</span>    visitBulkHFiles(fs, hfofDir, new BulkHFileVisitor&lt;byte[]&gt;() {<a name="line.306"></a>
+<span class="sourceLineNo">307</span>      @Override<a name="line.307"></a>
+<span class="sourceLineNo">308</span>      public byte[] bulkFamily(final byte[] familyName) {<a name="line.308"></a>
+<span class="sourceLineNo">309</span>        return familyName;<a name="line.309"></a>
+<span class="sourceLineNo">310</span>      }<a name="line.310"></a>
+<span class="sourceLineNo">311</span>      @Override<a name="line.311"></a>
+<span class="sourceLineNo">312</span>      public void bulkHFile(final byte[] family, final FileStatus hfile) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span>        long length = hfile.getLen();<a name="line.313"></a>
+<span class="sourceLineNo">314</span>        if (length &gt; getConf().getLong(HConstants.HREGION_MAX_FILESIZE,<a name="line.314"></a>
+<span class="sourceLineNo">315</span>            HConstants.DEFAULT_MAX_FILE_SIZE)) {<a name="line.315"></a>
+<span class="sourceLineNo">316</span>          LOG.warn("Trying to bulk load hfile " + hfile.getPath() + " with size: " +<a name="line.316"></a>
+<span class="sourceLineNo">317</span>              length + " bytes can be problematic as it may lead to oversplitting.");<a name="line.317"></a>
+<span class="sourceLineNo">318</span>        }<a name="line.318"></a>
+<span class="sourceLineNo">319</span>        ret.add(new LoadQueueItem(family, hfile.getPath()));<a name="line.319"></a>
+<span class="sourceLineNo">320</span>      }<a name="line.320"></a>
+<span class="sourceLineNo">321</span>    }, validateHFile);<a name="line.321"></a>
+<span class="sourceLineNo">322</span>  }<a name="line.322"></a>
+<span class="sourceLineNo">323</span><a name="line.323"></a>
+<span class="sourceLineNo">324</span>  /**<a name="line.324"></a>
+<span class="sourceLineNo">325</span>   * Perform a bulk load of the given directory into the given<a name="line.325"></a>
+<span class="sourceLineNo">326</span>   * pre-existing table.  This method is not threadsafe.<a name="line.326"></a>
+<span class="sourceLineNo">327</span>   *<a name="line.327"></a>
+<span class="sourceLineNo">328</span>   * @param hfofDir the directory that was provided as the output path<a name="line.328"></a>
+<span class="sourceLineNo">329</span>   *   of a job using HFileOutputFormat<a name="line.329"></a>
+<span class="sourceLineNo">330</span>   * @param admin the Admin<a name="line.330"></a>
+<span class="sourceLineNo">331</span>   * @param table the table to load into<a name="line.331"></a>
+<span class="sourceLineNo">332</span>   * @param regionLocator region locator<a name="line.332"></a>
+<span class="sourceLineNo">333</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.333"></a>
+<span class="sourceLineNo">334</span>   */<a name="line.334"></a>
+<span class="sourceLineNo">335</span>  public void doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.335"></a>
+<span class="sourceLineNo">336</span>      RegionLocator regionLocator) throws TableNotFoundException, IOException {<a name="line.336"></a>
+<span class="sourceLineNo">337</span>    doBulkLoad(hfofDir, admin, table, regionLocator, false, false);<a name="line.337"></a>
+<span class="sourceLineNo">338</span>  }<a name="line.338"></a>
+<span class="sourceLineNo">339</span><a name="line.339"></a>
+<span class="sourceLineNo">340</span>  void cleanup(Admin admin, Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.340"></a>
+<span class="sourceLineNo">341</span>      SecureBulkLoadClient secureClient) throws IOException {<a name="line.341"></a>
+<span class="sourceLineNo">342</span>    fsDelegationToken.releaseDelegationToken();<a name="line.342"></a>
+<span class="sourceLineNo">343</span>    if (bulkToken != null &amp;&amp; secureClient != null) {<a name="line.343"></a>
+<span class="sourceLineNo">344</span>      secureClient.cleanupBulkLoad(admin.getConnection(), bulkToken);<a name="line.344"></a>
+<span class="sourceLineNo">345</span>    }<a name="line.345"></a>
+<span class="sourceLineNo">346</span>    if (pool != null) {<a name="line.346"></a>
+<span class="sourceLineNo">347</span>      pool.shutdown();<a name="line.347"></a>
+<span class="sourceLineNo">348</span>    }<a name="line.348"></a>
+<span class="sourceLineNo">349</span>    if (!queue.isEmpty()) {<a name="line.349"></a>
+<span class="sourceLineNo">350</span>      StringBuilder err = new StringBuilder();<a name="line.350"></a>
+<span class="sourceLineNo">351</span>      err.append("-------------------------------------------------\n");<a name="line.351"></a>
+<span class="sourceLineNo">352</span>      err.append("Bulk load aborted with some files not yet loaded:\n");<a name="line.352"></a>
+<span class="sourceLineNo">353</span>      err.append("-------------------------------------------------\n");<a name="line.353"></a>
+<span class="sourceLineNo">354</span>      for (LoadQueueItem q : queue) {<a name="line.354"></a>
+<span class="sourceLineNo">355</span>        err.append("  ").append(q.hfilePath).append('\n');<a name="line.355"></a>
+<span class="sourceLineNo">356</span>      }<a name="line.356"></a>
+<span class="sourceLineNo">357</span>      LOG.error(err);<a name="line.357"></a>
+<span class="sourceLineNo">358</span>    }<a name="line.358"></a>
+<span class="sourceLineNo">359</span>  }<a name="line.359"></a>
+<span class="sourceLineNo">360</span>  /**<a name="line.360"></a>
+<span class="sourceLineNo">361</span>   * Perform a bulk load of the given directory into the given<a name="line.361"></a>
+<span class="sourceLineNo">362</span>   * pre-existing table.  This method is not threadsafe.<a name="line.362"></a>
+<span class="sourceLineNo">363</span>   *<a name="line.363"></a>
+<span class="sourceLineNo">364</span>   * @param map map of family to List of hfiles<a name="line.364"></a>
+<span class="sourceLineNo">365</span>   * @param admin the Admin<a name="line.365"></a>
+<span class="sourceLineNo">366</span>   * @param table the table to load into<a name="line.366"></a>
+<span class="sourceLineNo">367</span>   * @param regionLocator region locator<a name="line.367"></a>
+<span class="sourceLineNo">368</span>   * @param silence true to ignore unmatched column families<a name="line.368"></a>
+<span class="sourceLineNo">369</span>   * @param copyFile always copy hfiles if true<a name="line.369"></a>
+<span class="sourceLineNo">370</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.370"></a>
+<span class="sourceLineNo">371</span>   */<a name="line.371"></a>
+<span class="sourceLineNo">372</span>  public void doBulkLoad(Map&lt;byte[], List&lt;Path&gt;&gt; map, final Admin admin,<a name="line.372"></a>
+<span class="sourceLineNo">373</span>      Table table, RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.373"></a>
+<span class="sourceLineNo">374</span>          throws TableNotFoundException, IOException {<a name="line.374"></a>
+<span class="sourceLineNo">375</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.375"></a>
+<span class="sourceLineNo">376</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.376"></a>
+<span class="sourceLineNo">377</span>    }<a name="line.377"></a>
+<span class="sourceLineNo">378</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.378"></a>
+<span class="sourceLineNo">379</span>    // happen in this thread<a name="line.379"></a>
+<span class="sourceLineNo">380</span>    Deque&lt;LoadQueueItem&gt; queue = new LinkedList&lt;&gt;();<a name="line.380"></a>
+<span class="sourceLineNo">381</span>    ExecutorService pool = null;<a name="line.381"></a>
+<span class="sourceLineNo">382</span>    SecureBulkLoadClient secureClient = null;<a name="line.382"></a>
+<span class="sourceLineNo">383</span>    try {<a name="line.383"></a>
+<span class="sourceLineNo">384</span>      prepareHFileQueue(map, table, queue, silence);<a name="line.384"></a>
+<span class="sourceLineNo">385</span>      if (queue.isEmpty()) {<a name="line.385"></a>
+<span class="sourceLineNo">386</span>        LOG.warn("Bulk load operation did not get any files to load");<a name="line.386"></a>
+<span class="sourceLineNo">387</span>        return;<a name="line.387"></a>
+<span class="sourceLineNo">388</span>      }<a name="line.388"></a>
+<span class="sourceLineNo">389</span>      pool = createExecutorService();<a name="line.389"></a>
+<span class="sourceLineNo">390</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.390"></a>
+<span class="sourceLineNo">391</span>      for (Map.Entry&lt;byte[], List&lt;Path&gt;&gt; entry : map.entrySet()) {<a name="line.391"></a>
+<span class="sourceLineNo">392</span>        for (Path p : entry.getValue()) {<a name="line.392"></a>
+<span class="sourceLineNo">393</span>          fs = p.getFileSystem(table.getConfiguration());<a name="line.393"></a>
+<span class="sourceLineNo">394</span>          break;<a name="line.394"></a>
+<span class="sourceLineNo">395</span>        }<a name="line.395"></a>
+<span class="sourceLineNo">396</span>      }<a name="line.396"></a>
+<span class="sourceLineNo">397</span>      retValue = performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.397"></a>
+<span class="sourceLineNo">398</span>    } finally {<a name="line.398"></a>
+<span class="sourceLineNo">399</span>      cleanup(admin, queue, pool, secureClient);<a name="line.399"></a>
+<span class="sourceLineNo">400</span>    }<a name="line.400"></a>
+<span class="sourceLineNo">401</span>  }<a name="line.401"></a>
+<span class="sourceLineNo">402</span><a name="line.402"></a>
+<span class="sourceLineNo">403</span>  /**<a name="line.403"></a>
+<span class="sourceLineNo">404</span>   * Perform a bulk load of the given directory into the given<a name="line.404"></a>
+<span class="sourceLineNo">405</span>   * pre-existing table.  This method is not threadsafe.<a name="line.405"></a>
+<span class="sourceLineNo">406</span>   *<a name="line.406"></a>
+<span class="sourceLineNo">407</span>   * @param hfofDir the directory that was provided as the output path<a name="line.407"></a>
+<span class="sourceLineNo">408</span>   *   of a job using HFileOutputFormat<a name="line.408"></a>
+<span class="sourceLineNo">409</span>   * @param admin the Admin<a name="line.409"></a>
+<span class="sourceLineNo">410</span>   * @param table the table to load into<a name="line.410"></a>
+<span class="sourceLineNo">411</span>   * @param regionLocator region locator<a name="line.411"></a>
+<span class="sourceLineNo">412</span>   * @param silence true to ignore unmatched column families<a name="line.412"></a>
+<span class="sourceLineNo">413</span>   * @param copyFile always copy hfiles if true<a name="line.413"></a>
+<span class="sourceLineNo">414</span>   * @throws TableNotFoundException if table does not yet exist<a name="line.414"></a>
+<span class="sourceLineNo">415</span>   */<a name="line.415"></a>
+<span class="sourceLineNo">416</span>  public void doBulkLoad(Path hfofDir, final Admin admin, Table table,<a name="line.416"></a>
+<span class="sourceLineNo">417</span>      RegionLocator regionLocator, boolean silence, boolean copyFile)<a name="line.417"></a>
+<span class="sourceLineNo">418</span>          throws TableNotFoundException, IOException {<a name="line.418"></a>
+<span class="sourceLineNo">419</span>    if (!admin.isTableAvailable(regionLocator.getName())) {<a name="line.419"></a>
+<span class="sourceLineNo">420</span>      throw new TableNotFoundException("Table " + table.getName() + " is not currently available.");<a name="line.420"></a>
+<span class="sourceLineNo">421</span>    }<a name="line.421"></a>
+<span class="sourceLineNo">422</span><a name="line.422"></a>
+<span class="sourceLineNo">423</span>    /*<a name="line.423"></a>
+<span class="sourceLineNo">424</span>     * Checking hfile format is a time-consuming operation, we should have an option to skip<a name="line.424"></a>
+<span class="sourceLineNo">425</span>     * this step when bulkloading millions of HFiles. See HBASE-13985.<a name="line.425"></a>
+<span class="sourceLineNo">426</span>     */<a name="line.426"></a>
+<span class="sourceLineNo">427</span>    boolean validateHFile = getConf().getBoolean("hbase.loadincremental.validate.hfile", true);<a name="line.427"></a>
+<span class="sourceLineNo">428</span>    if (!validateHFile) {<a name="line.428"></a>
+<span class="sourceLineNo">429</span>      LOG.warn("You are skipping HFiles validation, it might cause some data loss if files " +<a name="line.429"></a>
+<span class="sourceLineNo">430</span>          "are not correct. If you fail to read data from your table after using this " +<a name="line.430"></a>
+<span class="sourceLineNo">431</span>          "option, consider removing the files and bulkload again without this option. " +<a name="line.431"></a>
+<span class="sourceLineNo">432</span>          "See HBASE-13985");<a name="line.432"></a>
+<span class="sourceLineNo">433</span>    }<a name="line.433"></a>
+<span class="sourceLineNo">434</span>    // LQI queue does not need to be threadsafe -- all operations on this queue<a name="line.434"></a>
+<span class="sourceLineNo">435</span>    // happen in this thread<a name="line.435"></a>
+<span class="sourceLineNo">436</span>    Deque&lt;LoadQueueItem&gt; queue = new LinkedList&lt;&gt;();<a name="line.436"></a>
+<span class="sourceLineNo">437</span>    ExecutorService pool = null;<a name="line.437"></a>
+<span class="sourceLineNo">438</span>    SecureBulkLoadClient secureClient = null;<a name="line.438"></a>
+<span class="sourceLineNo">439</span>    try {<a name="line.439"></a>
+<span class="sourceLineNo">440</span>      prepareHFileQueue(hfofDir, table, queue, validateHFile, silence);<a name="line.440"></a>
+<span class="sourceLineNo">441</span><a name="line.441"></a>
+<span class="sourceLineNo">442</span>      if (queue.isEmpty()) {<a name="line.442"></a>
+<span class="sourceLineNo">443</span>        LOG.warn("Bulk load operation did not find any files to load in " +<a name="line.443"></a>
+<span class="sourceLineNo">444</span>            "directory " + hfofDir != null ? hfofDir.toUri() : "" + ".  Does it contain files in " +<a name="line.444"></a>
+<span class="sourceLineNo">445</span>            "subdirectories that correspond to column family names?");<a name="line.445"></a>
+<span class="sourceLineNo">446</span>        return;<a name="line.446"></a>
+<span class="sourceLineNo">447</span>      }<a name="line.447"></a>
+<span class="sourceLineNo">448</span>      pool = createExecutorService();<a name="line.448"></a>
+<span class="sourceLineNo">449</span>      secureClient = new SecureBulkLoadClient(table.getConfiguration(), table);<a name="line.449"></a>
+<span class="sourceLineNo">450</span>      retValue = performBulkLoad(admin, table, regionLocator, queue, pool, secureClient, copyFile);<a name="line.450"></a>
+<span class="sourceLineNo">451</span>    } finally {<a name="line.451"></a>
+<span class="sourceLineNo">452</span>      cleanup(admin, queue, pool, secureClient);<a name="line.452"></a>
+<span class="sourceLineNo">453</span>    }<a name="line.453"></a>
+<span class="sourceLineNo">454</span>  }<a name="line.454"></a>
+<span class="sourceLineNo">455</span><a name="line.455"></a>
+<span class="sourceLineNo">456</span>  Map&lt;LoadQueueItem, ByteBuffer&gt; performBulkLoad(final Admin admin, Table table,<a name="line.456"></a>
+<span class="sourceLineNo">457</span>      RegionLocator regionLocator, Deque&lt;LoadQueueItem&gt; queue, ExecutorService pool,<a name="line.457"></a>
+<span class="sourceLineNo">458</span>      SecureBulkLoadClient secureClient, boolean copyFile) throws IOException {<a name="line.458"></a>
+<span class="sourceLineNo">459</span>    int count = 0;<a name="line.459"></a>
+<span class="sourceLineNo">460</span><a name="line.460"></a>
+<span class="sourceLineNo">461</span>    if(isSecureBulkLoadEndpointAvailable()) {<a name="line.461"></a>
+<span class="sourceLineNo">462</span>      LOG.warn("SecureBulkLoadEndpoint is deprecated. It will be removed in future releases.");<a name="line.462"></a>
+<span class="sourceLineNo">463</span>      LOG.warn("Secure bulk load has been integrated into HBase core.");<a name="line.463"></a>
+<span class="sourceLineNo">464</span>    }<a name="line.464"></a>
+<span class="sourceLineNo">465</span><a name="line.465"></a>
+<span class="sourceLineNo">466</span>    //If using secure bulk load, get source delegation token, and<a name="line.466"></a>
+<span class="sourceLineNo">467</span>    //prepare staging directory and token<a name="line.467"></a>
+<span class="sourceLineNo">468</span>    // fs is the source filesystem<a name="line.468"></a>
+<span class="sourceLineNo">469</span>    fsDelegationToken.acquireDelegationToken(fs);<a name="line.469"></a>
+<span class="sourceLineNo">470</span>    bulkToken = secureClient.prepareBulkLoad(admin.getConnection());<a name="line.470"></a>
+<span class="sourceLineNo">471</span>    Pair&lt;Multimap&lt;ByteBuffer, LoadQueueItem&gt;, Set&lt;String&gt;&gt; pair = null;<a name="line.471"></a>
+<span class="sourceLineNo">472</span><a name="line.472"></a>
+<span class="sourceLineNo">473</span>    Map&lt;LoadQueueItem, ByteBuffer&gt; item2RegionMap = new HashMap&lt;&gt;();<a name="line.473"></a>
+<span class="sourceLineNo">474</span>    // Assumes that region splits can happen while this occurs.<a name="line.474"></a>
+<span class="sourceLineNo">475</span>    while (!queue.isEmpty()) {<a name="line.475"></a>
+<span class="sourceLineNo">476</span>      // need to reload split keys each iteration.<a name="line.476"></a>
+<span class="sourceLineNo">477</span>      final Pair&lt;byte[][], byte[][]&gt; startEndKeys = regionLocator.getStartEndKeys();<a name="line.477"></a>
+<span class="sourceLineNo">478</span>      if (count != 0) {<a name="line.478"></a>
+<span class="sourceLineNo">479</span>        LOG.info("Split occurred while grouping HFiles, retry attempt " +<a name="line.479"></a>
+<span class="sourceLineNo">480</span>            + count + " with " + queue.size() + " files remaining to group or split");<a name="line.480"></a>
+<span class="sourceLineNo">481</span>      }<a name="line.481"></a>
+<span class="sourceLineNo">482</span><a name="line.482"></a>
+<span class="sourceLineNo">483</span>      int maxRetries = getConf().getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.483"></a>
+<span class="sourceLineNo">484</span>      maxRetries = Math.max(maxRetries, startEndKeys.getFirst().length + 1);<a name="line.484"></a>
+<span class="sourceLineNo">485</span>      if (maxRetries != 0 &amp;&amp; count &gt;= maxRetries) {<a name="line.485"></a>
+<span class="sourceLineNo">486</span>        throw new IOException("Retry attempted " + count +<a name="line.486"></a>
+<span class="sourceLineNo">487</span>            " times without completing, bailing out");<a name="line.487"></a>
+<span class="sourceLineNo">488</span>      }<a name="line.488"></a>
+<span class="sourceLineNo">489</span>      count++;<a name="line.489"></a>
+<span class="sourceLineNo">490</span><a name="line.490"></a>
+<span class="sourceLineNo">491</span>      // Using ByteBuffer for byte[] equality semantics<a name="line.491"></a>
+<span class="sourceLineNo">492</span>      pair = groupOrSplitPhase(table, pool, queue, startEndKeys);<a name="line.492"></a>
+<span class="sourceLineNo">493</span>      Multimap&lt;ByteBuffer, LoadQueueItem&gt; regionGroups = pair.getFirst();<a name="line.493"></a>
+<span class="sourceLineNo">494</span><a name="line.494"></a>
+<span class="sourceLineNo">495</span>      if (!checkHFilesCountPerRegionPerFamily(regionGroups)) {<a name="line.495"></a>
+<span class="sourceLineNo">496</span>        // Error is logged inside checkHFilesCountPerRegionPerFamily.<a name="line.496"></a>
+<span class="sourceLineNo">497</span>        throw new IOException("Trying to load more than " + maxFilesPerRegionPerFamily<a name="line.497"></a>
+<span class="sourceLineNo">498</span>            + " hfiles to one family of one region");<a name="line.498"></a>
+<span class="sourceLineNo">499</span>      }<a name="line.499"></a>
+<span class="sourceLineNo">500</span><a name="line.500"></a>
+<span class="sourceLineNo">501</span>      bulkLoadPhase(table, admin.getConnection(), pool, queue, regionGroups, copyFile,<a name="line.501"></a>
+<span class="sourceLineNo">502</span>          item2RegionMap);<a name="line.502"></a>
+<span class="sourceLineNo">503</span><a name="line.503"></a>
+<span class="sourceLineNo">504</span>      // NOTE: The next iteration's split / group could happen in parallel to<a name="line.504"></a>
+<span class="sourceLineNo">505</span>      // atomic bulkloads assuming that there are splits and no merges, and<a name="line.505"></a>
+<span class="sourceLineNo">506</span>      // that we can atomically pull out the groups we want to retry.<a name="line.506"></a>
+<span class="sourceLineNo">507</span>    }<a name="line.507"></a>
+<span class="sourceLineNo">508</span><a name="line.508"></a>
+<span class="sourceLineNo">509</span>    if (!queue.isEmpty()) {<a name="line.509"></a>
+<span class="sourceLineNo">510</span>      throw new RuntimeException("Bulk load aborted with some files not yet loaded."<a name="line.510"></a>
+<span class="sourceLineNo">511</span>        + "Please check log for more details.");<a name="line.511"></a>
+<span class="sourceLineNo">512</span>    }<a name="line.512"></a>
+<span class="sourceLineNo">513</span>    return item2RegionMap;<a name="line.513"></a>
+<span class="sourceLineNo">514</span>  }<a name="line.514"></a>
+<span class="sourceLineNo">515</span><a name="line.515"></a>
+<span class="sourceLineNo">516</span>  protected ClientServiceCallable&lt;byte[]&gt; buildClientServiceCallable(final Connection conn,<a name="line.516"></a>
+<span class="sourceLineNo">517</span>      TableName tableName, byte[] first, Collection&lt;LoadQueueItem&gt; lqis, boolean copyFile) {<a name="line.517"></a>
+<span class="sourceLineNo">518</span><a name="line.518"></a>
+<span class="sourceLineNo">519</span>    final List&lt;Pair&lt;byte[], String&gt;&gt; famPaths = new ArrayList&lt;&gt;(lqis.size());<a name="line.519"></a>
+<span class="sourceLineNo">520</span>    for (LoadQueueItem lqi : lqis) {<a name="line.520"></a>
+<span class="sourceLineNo">521</span>        famPaths.add(Pair.newPair(lqi.family, lqi.hfilePath.toString()));<a name="line.521"></a>
+<span class="sourceLineNo">522</span>    }<a name="line.522"></a>
+<span class="sourceLineNo">523</span><a name="line.523"></a>
+<span class="sourceLineNo">524</span>    return new ClientServiceCallable&lt;byte[]&gt;(conn,<a name="line.524"></a>
+<span class="sourceLineNo">525</span>        tableName, first, rpcControllerFactory.newController()) {<a name="line.525"></a>
+<span class="sourceLineNo">526</span>      @Override<a name="line.526"></a>
+<span class="sourceLineNo">527</span>      protected byte[] rpcCall() throws Exception {<a name="line.527"></a>
+<span class="sourceLineNo">528</span>        SecureBulkLoadClient secureClient = null;<a name="line.528"></a>
+<span class="sourceLineNo">529</span>        boolean success = false;<a name="line.529"></a>
+<span class="sourceLineNo">530</span>        try {<a name="line.530"></a>
+<span class="sourceLineNo">531</span>          LOG.debug("Going to connect to server " + getLocation() + " for row "<a name="line.531"></a>
+<span class="sourceLineNo">532</span>              + Bytes.toStringBinary(getRow()) + " with hfile group " +<a name="line.532"></a>
+<span class="sourceLineNo">533</span>              LoadIncrementalHFiles.this.toString( famPaths));<a name="line.533"></a>
 <span class="sourceLineNo">534</span>          byte[] regionName = getLocation().getRegionInfo().getRegionName();<a name="line.534"></a>
 <span class="sourceLineNo">535</span>          try (Table table = conn.getTable(getTableName())) {<a name="line.535"></a>
 <span class="sourceLineNo">536</span>            secureClient = new SecureBulkLoadClient(getConf(), table);<a name="line.536"></a>
@@ -638,7 +638,7 @@
 <span class="sourceLineNo">630</span>    ThreadFactoryBuilder builder = new ThreadFactoryBuilder();<a name="line.630"></a>
 <span class="sourceLineNo">631</span>    builder.setNameFormat("LoadIncrementalHFiles-%1$d");<a name="line.631"></a>
 <span class="sourceLin

<TRUNCATED>

Mime
View raw message